Skip to main content
DoneThat

AI Adoption GuideNonprofitOutreach

Agentic Campaign Sequence Builder

Agentic system produces a full multi-touch outreach campaign from an audience brief and channel preferences.

Nonprofit processPlanFundOutreachDeliverMeasureReportStewardRenew

By Don, DoneThat’s AI coach · updated

What this use case covers

An agentic campaign sequence builder turns a concise audience brief and a set of channel preferences into a complete multi-touch outreach plan. The output is a draft sequence: touch order, suggested timing, channel for each step, and working copy for email, mail, SMS, phone scripts, or other approved channels. It is not a live send engine and it does not replace your CRM’s scheduling or consent rules.

This page is for donor-communications and annual-fund leads who already know the audience and the ask, and who need a coherent sequence drafted quickly so staff can refine and approve it. The agent proposes; people decide what goes out.

Inputs the agent needs

The builder should not invent a campaign from a vague goal. It needs a usable brief and hard channel constraints.

Audience brief (minimum): who you are talking to (segment or list definition), why now (appeal, renewal, event, stewardship), the primary ask or action, any offer or matching gift context, tone and brand constraints, and exclusions (recent ask, do-not-contact, gift officers already in motion).

Channel preferences (minimum): which channels are allowed, preferred order or mix, hard caps (for example no SMS, email-only, or mail as the only first touch), quiet periods, and any compliance notes your organization treats as non-negotiable.

Optional but useful inputs include prior campaign performance notes, sample language that must be reused or avoided, gift officer talking points for phone steps, and links between this sequence and related stewardship or renewal tracks.

When the audience brief or channel constraints are missing, incomplete in a way that blocks a coherent plan, or contradictory (for example “SMS required” while channel preferences forbid SMS), the system returns empty output and surfaces what is missing. Empty output is the correct failure mode. A guessed sequence is worse than no sequence.

How the draft sequence is produced

Given a complete brief, the agent assembles a multi-touch outline first, then fills each touch with draft copy and timing suggestions.

Typical structure includes:

  1. Opening touch aligned to the channel mix (often email or mail for annual fund, sometimes a warm personal note for higher-touch segments).
  2. Follow-ups that advance the same narrative rather than restarting it: reminder of the ask, social proof or impact detail, deadline or matching gift window, and a final courtesy close.
  3. Channel handoffs where appropriate (email then phone, mail then email, SMS only where consented and allowed).
  4. Stop rules draft language: when a gift posts, when someone opts out, when a gift officer claims the household.

The agent should keep one through-line across touches: same audience, same primary ask, consistent facts. Variation belongs in angle and channel, not in changing the offer mid-sequence without staff intent.

Copy drafts should be labeled as drafts: subject lines, preview text, body, call scripts, and SMS character-safe variants when SMS is in scope. Timing suggestions are proposals (for example day-of-week windows or gaps between touches), not calendar commits, until someone maps them into your ESP or CRM journey tool.

What stays with staff

Human-in-the-loop is mandatory for this use case.

Staff still:

  • Approve or rewrite all copy before anything is loaded for send.
  • Approve cadence: number of touches, spacing, quiet periods, and whether a phone or mail step is warranted for this segment.
  • Confirm list membership, suppression, consent, and gift-officer ownership against the CRM.
  • Decide whether the sequence runs as a one-time campaign, a journey template, or a hybrid with personal follow-up.
  • Own brand, legal, and gift-acceptance language.

The agent accelerates drafting of a coherent plan. It does not auto-enroll donors, trigger sends, or override your approval workflow. Treat every generated sequence as a work product for review, the same way you would treat a junior writer’s first pass.

Where it fits in the outreach workflow

Use this after you know who and roughly why, and before you lock creative in the ESP.

A practical order of operations:

  1. Define or refresh the segment (behavior, affinity, giving history, or campaign eligibility).
  2. Score or flag lapse and priority where that informs urgency or channel intensity.
  3. Write the brief and channel constraints.
  4. Generate the sequence draft.
  5. Edit, approve, load into systems, and monitor.

The builder is especially useful when the same annual-fund story must be retargeted across several segments with different intensity: lapsed mid-level, recent first-time givers, loyal renewals, event attendees who have not yet given. Each brief yields its own sequence; staff keep a shared narrative spine so the organization does not sound like five different shops.

Quality checks before you accept a draft

Reject or regenerate a draft when:

  • Touches contradict each other on amount, deadline, or matching gift terms.
  • Channel mix violates the stated preferences or consent posture.
  • Timing ignores quiet periods, board blackouts, or overlapping appeals already scheduled.
  • Copy is generic enough that it could apply to any segment (sign that the brief was thin or ignored).
  • Phone or mail steps appear for audiences you never staff for those channels.
  • Stop rules are missing or weak relative to your CRM gift posting and preference center.

Accept a draft when the spine is clear, each touch has a job, channels match constraints, and editors can finish in a short review cycle rather than rewriting from scratch.

Is this worth automating for you?

Whether this pays back depends on how much time it takes your team today. Most teams estimate that from memory, and the estimate is usually wrong in one direction or the other. This one is rated high effort to implement, so the baseline matters more than usual.

DoneThat reconstructs where the time actually went, with no timers to forget, so you can measure the baseline before committing to a project and check the gain afterward.

Measure the baseline first